def testCactusSetup(self):
"""Creates a bunch of random inputs and then passes them to cactus setup.
"""
for test in xrange(self.testNo):
tempDir = os.path.relpath(getTempDirectory(os.getcwd()))
sequenceNumber = random.choice(xrange(100))
sequences, newickTreeString = getCactusInputs_random(
tempDir=tempDir, sequenceNumber=sequenceNumber)
#Setup the flower disk.
experiment = getCactusWorkflowExperimentForTest(
sequences, newickTreeString,
os.path.join('/data', os.path.relpath(tempDir)))
cactusDiskDatabaseString = experiment.getDiskDatabaseString()
cactusSequencesPath = os.path.join(experiment.getDbDir(),
"cactusSequences")
runCactusSetup(cactusDiskDatabaseString=cactusDiskDatabaseString,
cactusSequencesPath=cactusSequencesPath,
sequences=sequences,
newickTreeString=newickTreeString)
runCactusSetup(cactusDiskDatabaseString=cactusDiskDatabaseString,
cactusSequencesPath=cactusSequencesPath,
sequences=sequences,
newickTreeString=newickTreeString)
experiment.cleanupDb()
system("rm -rf %s" % tempDir)
logger.info("Finished test %i of cactus_setup.py", test)
def testCactusSetup(self):
"""Creates a bunch of random inputs and then passes them to cactus setup.
"""
for test in xrange(self.testNo):
tempDir = getTempDirectory(os.getcwd())
sequenceNumber = random.choice(xrange(100))
sequences, newickTreeString = getCactusInputs_random(tempDir=tempDir, sequenceNumber=sequenceNumber)
#Setup the flower disk.
experiment = getCactusWorkflowExperimentForTest(sequences, newickTreeString, tempDir)
cactusDiskDatabaseString = experiment.getDiskDatabaseString()
runCactusSetup(cactusDiskDatabaseString, sequences, newickTreeString)
runCactusSetup(cactusDiskDatabaseString, sequences, newickTreeString)
experiment.cleanupDb()
system("rm -rf %s" % tempDir)
logger.info("Finished test %i of cactus_setup.py", test)
def testCactus_Random_fixedAncestor(self):
"""Tests that cactus doesn't crash when aligning to a fixed ancestral sequence."""
sequences, _ = getCactusInputs_random(treeLeafNumber=3)
rootSeq = sequences.pop()
# Create a star tree
tree = '(%s)root;' % ",".join([str(x) + ":1.0" for x in range(len(sequences))])
outputDir = getTempDirectory()
experiment = getCactusWorkflowExperimentForTest(sequences, tree,
outputDir,
progressive=True)
experiment.setSequenceID("root", rootSeq)
experiment.setRootReconstructed(False)
experimentFile = os.path.join(outputDir, "experiment.xml")
experiment.writeXML(experimentFile)
jobTreeDir = os.path.join(outputDir, "jobTree")
self.progressiveFunction(experimentFile, jobTreeDir, 'singleMachine',
False, True, True, False)
def testCactusWorkflow_Blanchette(self):
"""Runs the workflow on blanchette's simulated (colinear) regions.
"""
if "SON_TRACE_DATASETS" not in os.environ:
return
for test in xrange(self.testNo):
tempFiles = []
tempDir = getTempDirectory(os.getcwd())
trueAlignment = os.path.join(TestStatus.getPathToDataSets(), "blanchettesSimulation", "00.job", "true.mfa")
#Load the true alignment.
columnAlignment = [ i for i in fastaAlignmentRead(trueAlignment) ]
fastaHeaders = [ i for i in fastaReadHeaders(trueAlignment) ]
sequenceNumber = 9
#The tree
newickTreeString = "((((HUMAN:0.006969, CHIMP:0.009727):0.025291, BABOON:0.044568):0.11,(RAT:0.072818, MOUSE:0.081244):0.260342):0.023260,((DOG:0.07, CAT:0.07):0.087381,(PIG:0.06, COW:0.06):0.104728):0.04);"
#Get random dir
testDir = getTempDirectory(tempDir)
#random alignment
alignmentLength = 5000
randomStart = random.choice(xrange(len(columnAlignment)-alignmentLength))
subAlignment = columnAlignment[randomStart:randomStart+alignmentLength]
logger.info("Got a sub alignment, it is %i columns long" % len(subAlignment))
#Get sequences
sequences = [ (fastaHeaders[seqNo], "".join([ column[seqNo] for column in subAlignment if column[seqNo] != '-' ])) for seqNo in xrange(sequenceNumber) ]
logger.info("Got the sequences")
#Write sequences into temp files
tempFastaFiles = []
for seqNo in xrange(sequenceNumber):
header, sequence = sequences[seqNo]
logger.info("Making temp file for header: %s, seq: %s" % (header, sequence))
tempFastaFile = os.path.join(testDir, "%i.fa" % seqNo)
tempFastaFiles.append(tempFastaFile)
fileHandle = open(tempFastaFile, "w")
fastaWrite(fileHandle, header, sequence)
fileHandle.close()
logger.info("Got the temp sequence files")
experiment = getCactusWorkflowExperimentForTest(tempFastaFiles, newickTreeString, testDir)
experimentFile = os.path.join(testDir, "experiment.xml")
experiment.writeXML(experimentFile)
cactusDiskDatabaseString = experiment.getDiskDatabaseString()
jobTree = os.path.join(testDir, "jobTree")
runCactusWorkflow(experimentFile, jobTree)
logger.info("Ran the the workflow")
#Check the output alignment
runJobTreeStatusAndFailIfNotComplete(jobTree)
logger.info("Checked the job tree dir")
#Output the 'TRUE' alignment file
if os.system("mfaToMaf --help > /dev/null 2>&1") == 0 and\
os.system("cactus_MAFGenerator --help > /dev/null 2>&1") == 0 and\
os.system("mafComparator --help > /dev/null 2>&1") == 0 and\
os.system("cactus_treeStats --help > /dev/null 2>&1") == 0:
trueMFAFile = os.path.join(testDir, "true.mfa")
fastaAlignmentWrite(subAlignment, fastaHeaders, len(fastaHeaders), trueMFAFile)
trueMAFFile = os.path.join(testDir, "true.maf")
system("mfaToMaf --mfaFile %s --outputFile %s --logLevel %s" % (trueMFAFile, trueMAFFile, getLogLevelString()))
system("cat %s" % trueMAFFile)
#Now get mafs for the region.
mAFFile = os.path.join(testDir, "flower.maf")
system("cactus_MAFGenerator --flowerName 0 --cactusDisk '%s' --outputFile %s --logLevel %s" % (cactusDiskDatabaseString, mAFFile, getLogLevelString()))
logger.info("Got the MAFs from the flower disk")
system("cat %s" % mAFFile)
statsFile = os.path.join(testDir, "stats.xml")
system("cactus_treeStats --cactusDisk '%s' --flowerName 0 --outputFile %s --logLevel %s" % (cactusDiskDatabaseString, statsFile, getLogLevelString()))
system("cat %s" % statsFile)
logger.info("Got the cactus tree stats")
#Now compare the mafs to the output.
resultsFile = os.path.join(testDir, "results.xml")
system("mafComparator --mafFile1 %s --mafFile2 %s --outputFile %s --logLevel %s" % (trueMAFFile, mAFFile, resultsFile, getLogLevelString()))
logger.info("Ran the maf comparator")
system("cat %s" % resultsFile)
#Cleanup
experiment.cleanupDb()
system("rm -rf %s" % testDir)
logger.info("Successfully ran test for the problem")
for tempFile in tempFiles:
os.remove(tempFile)
system("rm -rf %s" % tempDir)
